Develops, tests, and deploys healthcare AI models using clinical data and specialized medical coding systems.
PyHealth is a comprehensive Python library specifically designed for healthcare AI, offering a modular pipeline for processing electronic health records (EHR), physiological signals, and medical imaging. This skill enables developers to implement clinical prediction tasks like mortality and readmission, translate between medical coding systems (ICD, NDC, ATC), and utilize specialized deep learning architectures like RETAIN and SafeDrug. It streamlines the entire machine learning workflow from standardized data loading to clinical model evaluation, ensuring high-performance processing optimized for complex medical datasets.
主な機能
010 GitHub stars
02Standardized loading for 10+ major healthcare datasets including MIMIC, eICU, and OMOP
03Advanced medical coding translation between ICD-9/10, NDC, RxNorm, and ATC
04Clinical-grade evaluation tools for fairness, calibration, and uncertainty quantification
05Predefined clinical prediction tasks for mortality, readmission, and length of stay
06Library of 33+ models including healthcare-specific architectures like RETAIN and SafeDrug
ユースケース
01Developing safe medication recommendation systems with drug-drug interaction constraints
02Classifying sleep stages and detecting seizures from physiological EEG/ECG signals
03Predicting ICU patient mortality and hospital readmission risk using sequential EHR data
